You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
There are a couple of changes I would like to propose to the current approach for regression testing in OpenFast:
Currently regression tests in OpenFast check every entry in the binary out file. As a result we also check for very small values which may not be meaningful from an engineering standpoint. This approach requires anyone using OpenFast to have the exact hardware/software setup as the machine on which these results were generated. This hinders our attempt at truly being open-source. Consequently I propose not checking for post-processed variables which are irrelevant for a given regression test (e.g. RootFxr and RootFzr in bd_curved_beam).
There is a need for a more sophisticated and flexible test environment. Currently we test for every single entry in the binary out file. There are some tests that produce thousands of lines of output data and testing for every single one of them seems to be an overkill, and also a waste of time as the number of tests continue to increase. I foresee this being resolved via an additional file where the developer may specify various variables and the corresponding time steps at which they should be checked.
The text was updated successfully, but these errors were encountered:
@ashesh2512 can we make this issue "Improving robustness/flexibility in testing" in general? I have some thoughts on improvements to the work flow for development and testing which lie outside regression tests.
ashesh2512
changed the title
Improving robustness/flexibility of regression testing
Improving robustness/flexibility in testing
Jun 8, 2018
There are a couple of changes I would like to propose to the current approach for regression testing in OpenFast:
Currently regression tests in OpenFast check every entry in the binary out file. As a result we also check for very small values which may not be meaningful from an engineering standpoint. This approach requires anyone using OpenFast to have the exact hardware/software setup as the machine on which these results were generated. This hinders our attempt at truly being open-source. Consequently I propose not checking for post-processed variables which are irrelevant for a given regression test (e.g. RootFxr and RootFzr in bd_curved_beam).
There is a need for a more sophisticated and flexible test environment. Currently we test for every single entry in the binary out file. There are some tests that produce thousands of lines of output data and testing for every single one of them seems to be an overkill, and also a waste of time as the number of tests continue to increase. I foresee this being resolved via an additional file where the developer may specify various variables and the corresponding time steps at which they should be checked.
The text was updated successfully, but these errors were encountered: